Autophagy-mediated degradation of Fatty Acid Synthase (FASN) facilitates ATRA-induced granulocytic differentiation of acute myeloid leukemia (AML) cells

Acute myeloid leukemia (AML) is a blood cancer characterized by a block in differentiation and increased survival of the resulting AML blast cells. While standard chemotherapy for AML cures only 30% of patients, differentiation-inducing therapy using all-trans retinoic acid (ATRA) in combination with arsenic trioxide achieves 90% cure rates in acute promyelocytic leukemia (APL). A better understanding of the molecular mechanisms underlying ATRA therapy in APL may provide new perspectives in the treatment of additional AML subtypes. Fatty acid synthase (FASN) is the only human lipogenic enzyme available for de novo fatty acid synthesis. While FASN levels are very low in healthy adult tissues, it is often highly expressed in cancer cells, thus representing a potential therapeutic target. We found that FASN mRNA levels were significantly higher in AML patients than in healthy granulocytes or CD34+ hematopoietic progenitors in two AML cohorts (n=68, n=203) (p


Introduction
While traditional chemotherapy and radiotherapy aim to kill highly proliferative cancer cells, differentiation-inducing therapy aims to restore differentiation programs to drive cancer cells into maturation and ultimately into cell death. Differentiation therapies are associated with lower toxicity compared to classical cytotoxic therapies. The success of this therapeutic approach is exemplified by the introduction of all-trans retinoic acid (ATRA) in 1985 to treat acute promyelocytic leukemia (APL) 1 . The introduction of ATRA into the treatment regimen changed APL from being one of the most aggressive acute myeloid leukemia (AML) subtypes with a fatal course often within weeks only, to a curable disease with a complete remission rate of up to 95% of patients when combined with anthracycline-based chemotherapy or arsenic trioxide 1 . APL is characterized by translocations involving the C-terminus of the retinoic acid receptor alpha (RARA) on chromosome 17 and genes encoding for aggregate prone proteins.
Promyelocytic leukemia (PML)-RARA is the most frequently expressed fusion protein.
It is encoded by the translocation t(15;17) and has a dominant negative effect on RARA. RARA transcriptionally regulates multiple biological processes with a key role in differentiation 2 . Several reports suggest a beneficial effect of ATRA in combination therapies in non-APL AML cells [3][4][5] . Unfortunately, a variety of intrinsic resistance mechanisms in non-APL AML have been identified such as SCL overexpression, expression of PRAME and epigenetic silencing or mutation of RARA 6-9 . Deciphering the mechanisms active during ATRA-mediated differentiation at the molecular level will support the translation of differentiation therapy to non-APL AML patients. We and others have demonstrated the important role of autophagy in ATRA induced granulocytic differentiation of APL cells [10][11][12][13][14][15][16] . Autophagy is an intracellular degradation mechanism that ensures dynamic recycling of various cytoplasmic contents 17 . We thus BioRXiv aim to understand the role of autophagy in granulocytic differentiation and define key druggable autophagy targets in this process.
Endogenous synthesis of fatty acids is catalyzed by fatty acid synthase (FASN), the only human lipogenic enzyme able to perform de novo synthesis of fatty acids 18,19 .
FASN is frequently overexpressed in a variety of tumor types including leukemias [20][21][22][23][24][25][26] while its expression in healthy adult tissues is low 27 , with the exception of the cycling endometrium 28 and lactating breast 29 . Interestingly, FASN is upregulated in tumor associated myeloid cells where it activates nuclear receptor peroxisome-proliferatoractivated receptor beta/delta (PPARβ/δ) 30 , a key metabolic transcription factor in tumorigenesis 31,32 . Of note, activation of PPARβ/δ regulates anti-inflammatory phenotypes of myeloid cells in other biological contexts such as atherosclerosis and obesity [33][34][35][36] . We previously reported that (-)-epigallocatechin-3-gallate (EGCG) improved ATRA induced differentiation of APL cells by increasing the expression of death associated protein kinase 2 (DAPK2). Furthermore, EGCG treatment reduces FASN expression levels in selected breast cancer cell lines 37 . The increased FASN expression in cancer including leukemias, its function in tumor-associated myeloid cells and its link to the differentiation enhancer DAPK2 prompted us to analyze the regulation and function of FASN during leukemic differentiation.
In the present study, we demonstrate that FASN expression is significantly higher in AML blasts partially due to low autophagic activity in those cells. We show that inhibiting FASN protein expression, but not enzymatic activity, promotes differentiation of non-APL AML cells. Lastly, we link FASN expression to the inhibition of the key lysosomal biogenesis transcription factor TFEB.

Primary cells, cell lines and culture conditions
Fresh leukemic blast cells from untreated AML patients at diagnosis obtained at the Inselspital Bern (Switzerland) were classified according to the French-American-British classification and cytogenetic analysis. All leukemia samples had blast counts 90% after separation of mononuclear cells using a Ficoll gradient (Lymphoprep; Axon Lab AG, Switzerland), as described previously 38 . Protocols and use of 67 human samples acquired in Bern were approved by the Cantonal Ethical Committee at the Inselspital.
The isolation of primary neutrophils (purity 95%) was performed by separating blood cells from healthy donors using polymorphprep (AxonLab AG). CD34 + cells from cord blood or bone marrow were isolated as described 38 .

Cell lysate preparation and western blotting
Whole cell extracts were prepared using UREA lysis buffer and 30-60μg of total protein was loaded on a 7.5% or 12% denaturing polyacrylamide self-cast gel (Bio-Rad). Blots were incubated with the primary antibodies in TBS 0.05% Tween-20 / 5% milk overnight at 4°C and subsequently incubated with HRP coupled secondary goat antirabbit (7074; Cell signaling) and goat anti-mouse antibody (7076; Cell signaling) at 1:5-10,000 for 1 h at room temperature. Blots were imaged using the Chemidoc (Bio-Rad) and ImageLab software.
Lentivirus production and transduction were done as described 39,40 . Transduced NB4 cell populations were selected with 1.5 µg/ml puromycin for 4 days and knockdown efficiency was assessed by western blot analysis.

Immunofluorescence microscopy
BioRXiv Cells were prepared as previously described 15 . Briefly, cells were fixed and permeabilized with ice-cold 100% methanol for 4 min (LC3B and LAMP1 staining) or 2% paraformaldehyde for 7minutes followed by 5 minutes in PBS TRITONX (TFEB and tubulin staining) and then washed with PBS. Cells were incubated with primary antibody for 1 h at room temperature followed by washing steps with PBS containing 0.1% Tween (PBS-T). Cells were incubated with the secondary antibody (anti-rabbit, Acridine Orange staining was measured by FACS using Laser 488nm with 530/30 (GREEN) and filter 695/40 (RED) filters on a FACS LSR-II (BD). Data were analyzed with FlowJo software. The software derived the RED/GREEN ratio and we compared the distribution of populations using the Overton cumulative histogram subtraction algorithm to provide the percentage of cells more positive than the control.

Nitroblue tetrazolium reduction test
Suspension cells (5 x 10 5 ) were resuspended in a 0.2% nitro blue tetrazolium (NBT) solution containing 40ng/ml PMA and incubated 15min at 37°C. Cells were then BioRXiv washed with PBS, and subjected to cytospin. Counterstaining was done with 0.5% Safranin O for 5min (HT90432; Sigma). The NBT-positive and negative cells were scored under a light microscope (EVOS).

Trypan blue exclusion counting
Trypan blue exclusion cell counting was performed to assess cellular growth. 20µL of cells suspension was incubated with an equal volume of 0.4% (w/v) trypan blue solution (Sigma-Aldrich). Cells were counted using a dual-chamber hemocytometer and a light microscope.

Statistical analysis
Nonparametric Mann-Whitney-U tests were applied to compare the difference between two groups and Spearman Coefficient Correlation using Prism software. P-values < 0.05 were considered statistically significant.

Primary AML blast cells express significantly higher FASN levels compared to mature granulocytes
Cancer cells frequently express high levels of FASN compared to their healthy counterparts 20-26 . We examined FASN mRNA expression in an AML patient cohort with well-defined molecular subtypes. FASN mRNA levels in AML samples (n=68) were compared to the levels in granulocytes (n=5) and CD34 + human hematopoietic progenitor cells (n=3) from healthy donors. We found that FASN expression was significantly higher in AML patients compared to granulocytes and CD34 + hematopoietic progenitors from healthy donors (p<0.05) ( Figure 1A). We obtained similar findings by analyzing FASN expression in AML patient data available from the Bloodspot gene expression profile data base 41 ( Figure 1B). Of note, there were no significant differences in FASN expression in different AML subtypes (data not shown).
Next, we asked if FASN expression was altered during granulocytic differentiation of APL cells. We analyzed FASN expression following ATRA-induced differentiation of two APL cell lines, NB4 and HT93. ATRA treatment resulted in markedly reduced FASN protein levels from day two onwards ( Figure 1C). This further suggests that high FASN expression is linked to an immature blast-like phenotype and that ATRA-induced differentiation reduces the levels of FASN ( Figure 1C).

FASN protein is degraded via macroautophagy during ATRA-induced granulocytic differentiation
We and others have demonstrated that autophagy gene expression is repressed in AML samples compared to granulocytes from healthy donors and that autophagy activity is essential for successful ATRA-induced APL differentiation 10-16,42-44 . The decrease in FASN expression upon ATRA-induced differentiation cannot be explained solely by transcriptional regulation due to the long half-life of this protein (1-3 days) 45,46 . BioRXiv Moreover, FASN can be present inside autophogosomes, for instance in yeast and in the breast cancer cell MCF7 47,48 . Therefore, we hypothesized that ATRA-induced autophagy participates in the degradation of FASN during differentiation of APL cells.
To examine whether autophagy is important for degradation of FASN, we treated NB4 cells for 24h with different concentrations of Bafilomycin A1 (BafA1), a specific inhibitor of vacuolar-type H + -ATPase 49,50 , alone or in combination with ATRA. FASN protein was found to accumulate in the presence of BafA1, together with autophagy markers p62 and LC3B-II ( Figure 2A). To access if FASN degradation is feature of canonical autophagy, we assessed expression during starvation-induced autophagy. NB4 cells were starved for 2 hours using EBSS medium, a time period that induces autophagy without cell death (Supplementary Figure 1). Upon starvation, we detected lower FASN protein levels ( Figure 2B). Furthermore, FASN protein levels were slightly stabilized by inhibiting autolysosome formation using BafA1 (200nM) during starvation. This data suggests that FASN may be degraded via canonical autophagy ( Figure 2B).
To further assess this possibility, we utilized NB4 cells stably expressing mCherry-LC3B. Cells were treated with different concentrations of BafA1 with or without ATRA for 24h and FASN as well as LC3B localization was assessed. Endogenous FASN (cyan) showed co-localization with mCherry-LC3B (red) in BafA1 and ATRA treated cells ( Figure 2C, right panels). In addition, we found colocalization with endogenous FASN (red) and p62 (green) in NB4 parental cells treated with both ATRA and BafA1 for 24h ( Figure 2D). It is possible that p62 may help to sequester FASN to the autophagosome -although co-localisation of p62 to the lysosome would be expected.
In summary, this data suggests that FASN is a target for autophagic degradation during granulocytic differentiation of APL cells.

BioRXiv
We have previously shown that EGCG improves the response to ATRA in AML cells by inducing DAPK2 expression, a key kinase in granulocytic differentiation 51 .
Furthermore, EGCG was reported to decrease FASN expression 37 and this was reproducible in our APL cell line model (Supplementary Figure 2A In addition, by co-treating NB4 parental cells with EGCG and ATRA for 24h and increasing concentrations of BafA1, we clearly demonstrate that EGCG potentiates ATRA induced autophagy, leading to an increase in FASN protein degradation as confirmed by western blotting (Figure 3). Together our data demonstrate that FASN is degraded via autophagy upon ATRA treatment in APL cell lines and that co-treatment with EGCG promotes FASN protein degradation and increases its ubiquitination.

Inhibiting FASN protein expression but not its catalytic function leads to acceleration of ATRA-induced differentiation
Next, we evaluated the impact of modulating FASN expression and activity on myeloid differentiation. Therefore, we genetically inhibited FASN expression using lentiviral vectors expressing 2 independent shRNAs targeting FASN in the NB4 APL cell line model. Knockdown efficiency was validated by western blotting ( Figure 4A). We found  Figure 4G and Figure 4J). Therefore, we hypothesize that the catalytic activity of FASN is not involved in impeding ATRA-mediated differentiation in NB4 cells.

FASN is a negative regulator of autophagy by increasing mTOR activity
FASN has been previously reported to promote carcinogenesis by activating mTOR, a master negative regulator of autophagy, via AKT in hepatocellular carcinoma 54,55 .
ATRA treatment in APL also reduces mTOR activity leading to autophagy activation 56 .
We therefore hypothesized that FASN may negatively regulate autophagy via mTOR in APL cells, thereby impeding ATRA-induced differentiation. Therefore, we initially confirmed that FASN expression impacts on autophagic activity in our system. Autophagy induction was determined by quantifying endogenous, lipidated LC3B-II by BioRXiv western blotting and immunofluorescence microscopy (IF) after ATRA treatment 53 . In order to measure autophagic flux, ATRA treatment was performed in the presence or absence of BafA1 53 . Interestingly, although the autophagy flux (BafA1 + -BafA1 -) was not significantly increased based on LC3B-II measurement ( Figure 5 A-C), inhibition of FASN protein levels led to a drastic accumulation of p62 upon BafA1 treatment ( Figure   5C). Suggesting that FASN expression negatively regulates p62 expression or stability.
As LC3B levels are similarly enhanced -it would be consistent with elevation of autophagosomal incorporation of p62 -without lysosomal fusion to degrade it. ULK1 (ATG1), a key autophagy gene of the initiation complex, is inhibited by mTORmediated phosphorylation at Ser757, leading to reduced autophagic activity 57 . In line with FASN activating mTOR, lowering FASN protein expression by shRNA resulted in decreased mTOR-mediated phosphorylation of ULK1 at Ser757. Elevated ULK1 activity was confirmed by an increase of ATG13 phosphorylation at Ser318 mediated by active ULK1 58,59 . These results suggest that FASN expression promotes mTOR activity in AML cells, thereby enhancing its repression of autophagy.

FASN expression negatively affects transcription factor EB (TFEB)
activation mTOR regulates the transcription factor EB (TFEB), a master regulator of lysosome biogenesis) 60,61 . mTOR phosphorylates TFEB, leading to the sequestration of TFEB within the cytoplasm and inhibition of its transcriptional activity 62,63 . TFEB is a key transcriptional regulator of more than 500 genes that comprise the CLEAR Furthermore, we treated cells with Acridine Orange to quantify the lysosomal integrity by flow cytometry. Acridine Orange is a cell permeable fluorescent dye that, when excited at 488nm, emits light at 530nm (GREEN) in its monomeric form but shifts its emission to 680nm (RED) when accumulating and precipitating inside lysosomes. Therefore, we measured the RED/GREEN ratio of Acridine Orange stained cells by flow cytometry as previously described 65 . We found that ATRA treatment shifted the ratio towards the red channel (Supplementary Figure 6C). Reducing FASN expression further accelerated the increase of RED/GREEN ratio indicating enhanced lysosome biogenesis ( Figure 6 G-H).

BioRXiv
These results suggest that FASN expression impairs TFEB translocation to the nucleus and therefore reduces lysosome biogenesis. Next, we tested whether we can obtain similar results by treating the cells with EGCG instead of knocking down FASN.
Using different EGCG concentrations, we found a decrease in mTOR phosphorylation Together, these data suggest that high FASN expression results in lower autophagic activity and decreased lysosomal capacity due to increased mTOR activity and inhibition of TFEB.

Lowering FASN expression improves ATRA therapy in non-APL AML cell lines by inhibiting the mTOR pathway
Given the fact that APL cells treated with EGCG demonstrated improved response to ATRA therapy, we asked if EGCG can be beneficial to other AML subtypes that are refractory to ATRA treatment. We and others previously demonstrated a positive impact of co-treating HL60 AML cells, a non-APL AML cell line that responds to ATRA, with EGCG and ATRA. Therefore, we tested if additional ATRA refractory AML cell Together, this data suggests that reducing FASN expression can increase lysosomal biogenesis and improve the differentiation of non-APL AML cells.

Discussion
In this study, we aimed to further dissect the function of fatty acid synthase in AML cells and in particular, its potential role in the differentiation of immature blasts. We showed that knocking down FASN accelerated ATRA-induced differentiation, while inhibition of its enzymatic function by pharmacological inhibitors such as C75 or Orlistat had no effect. Furthermore, we found that FASN expression activates mTOR resulting

BioRXiv
We further confirmed that EGCG positively impacts cellular differentiation in additional AML subtypes in vitro 51, 70,71 . Searching for potential mediators of the positive effects of EGCG observed during ATRA-induced differentiation, we previously found that EGCG induces expression of the Ca 2+ /calmodulin-regulated serine/threonine kinase DAPK2.
DAPK2 plays a major role in granulocytic differentiation and decreased DAPK2 expression in APL cells can be restored by ATRA and EGCG treatment 14,39,51,72 .
DAPK2 also negatively regulates mTOR via phosphorylation of raptor at Ser721 as shown in HeLa cells 73 . Therefore, it would be of interest to study the impact of FASN on DAPK2 activity in a leukemic context as well.
Interestingly, treating APL cells with ATRA had a negative effect on FASN protein levels ( Figure 1C) and we demonstrated that ATRA-induced autophagy contributes to FASN protein degradation. Furthermore, FASN reduction led to increased lysosomal biogenesis suggesting a negative feedback loop between autophagy and FASN. It is reasonable to hypothesize that the more AML cells differentiate the more they become competent to degrade long-lived proteins including FASN. We and others demonstrated the importance of functional autophagy pathways during ATRA-induced differentiation 10-16,42-44 . In addition inhibiting mTOR using Rapamycin or Everolimus accelerates differentiation of APL cells 13,56 . While PI3K/Akt/mTOR pathways are activated in about 80% of AML cases, mTOR inhibitors had only modest effects in AML therapy 74,75 . Furthermore, despite its role in leukemia cells, mTOR activity is crucial for hematopoietic stem cell proliferation and self-renewal potential 76 . Therefore, targeting FASN that is low expressed in healthy cells would allow activation of autophagy in AML cells, resulting in an increased degradation of long-lived proteins, sparing healthy cells in the bone marrow. EGCG in our hands demonstrated a partial effect compared to the knock down. Therefore, a more specific FASN expression inhibitor will help to improve differentiation therapy in non-APL AML patients. In summary, our data suggest that inducing FASN protein degradation is likely to be beneficial for differentiation therapy of non-APL AML cells as this will impede mTOR and promote TFEB transcriptional activity and autophagy. Furthermore, high FASN expression in AML is likely due to impaired autophagy activity.

Acknowledgments
Deborah Shan-Krauer is gratefully acknowledged for excellent technical support. We